linguistic theory and the analysis of minority languages: native

Linguistic Theory and the Analysisof Minority Languages:Native, Immigrant and HeritageSpanish

Juana M. LicerasCristina Senn

In this paper we aim to contribute to the emerging field of heritagestudies by investigating whether Spanish heritage speakers inCanada, namely the second or subsequent generation of Spanishspeakers who grew up as English-Spanish bilinguals, differ fromnative Spanish speakers (those who have always lived in a Spanish-speaking country) and from immigrant Spanish speakers (those whoimmigrated to Canada as adults) with respect to their grammaticalcompetence and to their processing strategies. Taking as a point ofdeparture recent proposals from linguistic theory, we provide adescription of Spanish restrictive relative clauses with so-calledresumptive pronouns (Es una mujer que nunca LA vimos llorar) inorder to determine whether and how our three groups of speakersdiffer in terms of the grammatical intuitions and processing resourcesthey display when confronted with this type of constructions. Wediscuss to what extent language attrition, influence from English (inthe case of both immigrant and heritage speakers), or incompleteacquisition (in the case of heritage speakers) may be behind thecharacteristics of the immigrant and the heritage speakers’ linguisticbehaviour. We argue that sophisticated experimental tasks provide abetter tool than global proficiency tests to compare these threegroups of speakers. The ultimate aim of this study is to provide aframework for analyzing the status of the minority languages spokenby immigrant communities.

Este artículo quiere contribuir al emergente campo de los estudios dela herencia (heritage), investigando si los hablantes de herencia deespañol en Canadá (los hablantes de español de segunda o posteriorgeneración que crecieron como bilingües de inglés-español) sediferencian de los hablantes de español nativos (aquellos que hanvivido siempre en un país de habla hispana) y de los inmigrantes

39Lengua y migración 1:1 (2009), 39-73ISSN: 1889-5425. © Universidad de Alcalá

41

Juana M. Liceras y Cristina Senn

40

Linguistic theory and the analysis of minority languages

hablantes de español (los que inmigraron a Canadá como adultos),respecto a su capacidad gramatical y a sus estrategias deprocedimiento. Tomando como punto de partida las propuestas másrecientes de la teoría lingüística, presentamos una descripción de lascláusulas relativas restrictivas con pronombres de reasuntivos enespañol (Es una mujer que nunca LA vimos llorar) con el fin dedeterminar las diferencias entre los tres grupos de hablantes entérminos de intuiciones gramaticales y de recursos de procesamientoque utilizan al enfrentarse a este tipo de construcciones. Se discute enqué medida la erosión de la lengua, por influencia del inglés (en el casode los hablantes inmigrantes y de herencia), o una adquisiciónincompleta (en el caso de los hablantes de herencia) pueden estardetrás del comportamiento lingüístico. Se discute también si laspruebas experimentales sofisticadas son mejores herramientas, paracomparar los tres grupos de hablantes, que los tests de dominioglobal. El objetivo final de este estudio es proporcionar un marcopara el análisis del estatus de las lenguas minoritarias habladas por lascomunidades de inmigrantes.

Keywords: Heritage speakers - Immigrant speakers - Late contactbilinguals - Early contact bilinguals - Grammatical competence -Processing abilities - Ambiguity resolution –Resumptive pronouns –“Near native speaker” –– Acceptability Judgments — Languageattrition – Incomplete acquisition – Interlinguistic influence –Minority language.

1. IntroductionThere are at least two ways of conceptualizing the relationship betweenlanguage and immigration. One, which seems to be of great interest incountries such as Spain where immigration is a somehow recentphenomenon, addresses the problems encountered by the immigrantwhen learning the official language of the country—his or her second (orthird or fourth language)—. The other which, to the best of ourknowledge, is rather secondary in Spain right now but object to asubstantial amount of research in the United States (not so much inCanada), addresses the problems encountered by so-called heritagespeakers.

What is probably a clear example of a heritage speaker is the off-spring of immigrants. In other words, the speaker who has been exposedto a non-official language (therefore a minority language) in the houseand/or the community where she grows up and, at the same time, learnsthe official language, either simultaneously or before the age of three orfour, when she enters kindergarten. In this specific case, contact with the

minority language continues through the speaker’s life span but notnecessarily in the regular and consistent way that characterizes contactwith an official or majority language. Therefore, issues of quantity andquality of input will play an important role when it comes to languageacquisition by heritage speakers. In relation to this, attrition versusincomplete acquisition will have to be measured up against the fact thatthe input that we can refer to as standard may and probably will havebeen modified by the immigrant community. A variable that should alsobe taken into consideration when carrying out experimental work isliteracy, since there may be heritage speakers who are never schooled inthe minority language. This is very important when comparing secondlanguage learners with heritage speakers, since the former’s only contactmay be institutional and therefore, surrounded by a high level of literacy.In fact, as Pires and Rothman (2009) point out, “the sociolinguisticcircumstances of heritage language acquisition, access to and level offormal education in the heritage language, and input quantity and qualityavailable to HSs are some of the external variables that can conspire tomake HS language acquisition and its ensuing competence outcomesdifferent from both monolingual and balanced bilingual childhoodacquisition” (Pires and Rothman 2009:3).

The relationship between language and immigration that we want toaddress here is that of the first language or mother tongue of theimmigrant speaker and the second generation or third generationspeaker, the heritage speaker. The specific contact that we investigate isthat of Spanish immigrants who have been in contact with English, theirsecond language, for more than ten years (we will refer to this group asLate Contact—with English—group or immigrant group) and thatSpanish heritage speakers, in our case the off-springs of thoseimmigrants, the Early Contact—with English—group whose firstcontact with English took place at birth or before age four or five. To thebest of our knowledge, this comparison of immigrant and heritagespeakers has not been addressed before, which adds one more innovativeaspect to our contribution. It is not necessarily the case that heritagespeakers are only exposed to the language of the immigrant population,since they may travel to Spanish-speaking countries or have contact withfamily members who live in Spanish-speaking countries (our nativegroup). In this respect, if the immigrant speech differs from the nativespeech, it can be easily inferred that a substantial amount of the Spanishinput heritage speakers have been exposed to is different, in somerespects, from the native input they would be exposed to in the Spanish-speaking countries. However, it may not be input as such but the factthat it is a minority language and therefore it will only be “used and

heard in restrictive environments” (Rothman 2007: 360) what willaccount for the specific characteristics of the heritage language.

We have analyzed the linguistic behaviour of three groups ofspeakers and two varieties of Spanish: Castilian and Rioplantense, thereason being that the Rioplatense variety has some very well-knowndifferences with respect to the use of non-tonic pronouns (or clitics).However, it is other aspects of clitics such as clitic doubling that wouldshow differences. Therefore, in this article we will not be addressing thepossible differences between the two varieties of Spanish.1 The nativegroup, or Spanish control group (SC), is formed by twenty adult Spanishspeakers, ten from Spain and ten from Argentina who live in thesecountries and have not lived in a non-Spanish-speaking country orcommunity. For these speakers Spanish is clearly the dominant language.We refrain from calling the members of this control group monolingualsbecause they are educated speakers who have been in contact withEnglish and sometimes with other languages in their school years.

The Late Contact group (LC) comprises twelve immigrants whowere born in Argentina or Spain and who have been in Canada for atleast ten years. Their systematic contact with English, the majority andofficial language, begun late in their lives in that they were already 18years old or older when they immigrated to Canada and did have noneor little knowledge of English before their arrival. This group’s Spanishmay have suffered from attrition or their Spanish may have changed withrespect to the standard variety which is spoken by the inhabitants ofSpain and Argentina. However, if this change has occurred it will not bebecause a Canadian-Spanish variety of the language has emerged, at leastnot in the case of our subjects, since their situation is very different fromthat of the Spanish-speaking communities in the United States ofAmerica. Our subjects are professionals who maintain a good rapportwith other Spanish-speaking immigrants but do not have a tight networkor rely on the Spanish-speaking community for their social orprofessional life.

In the Early Contact (EC) group we have placed twelve participantsin total. Seven of these participants were born in Canada within aSpanish-speaking family and were in contact with both Spanish andEnglish from birth. Three were born in Spain or Argentina but came toCanada during their first year of age. Finally, two participants came toCanada when they were four years old. Therefore, all the members ofthis group were exposed to both Spanish and English well before the so-called critical period of language acquisition (Penfield 1953; Penfield andRoberts 1959; Lenneberg 1967; Johnson and Newport 1991, amongothers). These are the type of speakers that are referred to as heritage interms of their contact with Spanish. 4342

Linguistic theory and the analysis of minority languages Juana M. Liceras y Cristina Senn

We show that the two experimental groups (the immigrant and theheritage group) significantly differ from the control group in their globalcompetence as measured by a proficiency test. We also show that, in aself-evaluation task, the heritage group shows a higher degree ofinsecurity with respect to their knowledge of Spanish than theimmigrant group. This higher degree on insecurity correlates with thislatter group’s reduced sensibility to the grammatical gender depicted bythe resumptive pronouns and to the [-definite] [+distance] factors whichimprove the acceptability of these type of pronouns.

With respect to influence from English, we argue that it may accountfor the lower acceptance of resumptives by the immigrant group and theheritage group’s lesser sensitivity to the gender feature of the relativizedphrases. As for whether attrition or incomplete acquisition wouldaccount for the grammatical and the processing differences displayed bythe heritage group, we argue that it is necessary to differentiate not onlybetween grammatical and processing capabilities but also between thecore computational aspects involving relatives and clitic pronouns, onthe one hand, and the fuzzy status of native speakers intuitions withrespect to constraints on the use and acceptance of resumptive pronouns,on the other.

2. The grammar of heritage speakersIt has been shown (Silva-Corvalán 1994) that heritage speakers who hada high overall competence in their minority language (Spanish) also hada robust knowledge of the syntax of dative and accusative clitics in thislanguage. However, this was not the case when their overall competencewas reduced. In this latter case, they displayed a tendency to avoid cliticdoubling with indirect objects and had incomplete knowledge of thesemantics and pragmatics of both subjects and objects. In other words,in spite of the fact that these heritage speakers can speak the languagefluently, have a native-like pronunciation and have mastered thelanguage cultural norm, they have problems with some aspect of thesyntax and the morphology of the minority language. This implies that,as it has been shown in the case of fossilization (Selinker 1972, Liceras1986), incomplete acquisition or attrition also seems to be local andselective (Montrul 2008).

While heritage speakers have different levels of proficiency in theheritage language, many possess a rather good command of itsphonology, a fact that has been attributed to their being exposed to thelanguage at a very early age. In this respect, they are very different fromadult second language learners who, in spite of the fact that they mayachieve native-like competence in terms of the morpho-syntax

(Birdsong 1999, Lardiere 2007), seldom achieve native-like phonologicalcompetence. In terms of morphosyntax, some researchers claim that,unlike the case with the phonology, early contact with the heritagelanguage does not produce similar benefits (Oh et al. 2003), while others(Montrul 2006) argue that heritage speakers outperform secondlanguage learners of similar age and proficiency with respect to somespecific morphosyntactic aspects of the minority language.

Assumming that our immigrant or LC group is representative of thespeakers who provide the relevant input that leads our heritage group tobuild the grammar of Spanish, the first issue that we would like toaddress is whether our LC group global and local linguistic behaviourdiffers from the SC group in such a way that one could speak of theexistence of a different variety of Spanish. As we indicated in theintroduction, the membes of our EC group have not been inmersed in asociolinguistic Spanish-speaking community such as the Puerto Rican orthe Mexican communities in the large cities of the United States. Thosewhose parents speak the Castilian variety have been mostly in contactwith other Spanish-speaking families and the same can be said about thegroup whose parents speak the Rioplatense variety of Spanish. Thus, byanalyzing the linguistic behaviour of both the LC and EC groups wewill be able to determine whether, besides systematicity and quantity ofinput, quality of input can also account for the specific characteristics ofheritage Spanish.

The second issue that we would like to address is literacy in Spanishand how it could be one of the contributing factors in determining thespecific characteristics of heritage Spanish (Rothman 2007). In thisrespect, we believe that we have factored out literacy by investigating oursubjects intuitions and processing strategies with respect to resumptivepronouns, since these constructions as such are neither discussed in thenative Spanish nor in the foreign Spanish classroom. In other words, aswe will see, it seems to be sensitivity to the realization of the principlesof the computational component of Spanish that leads native speakers tobehave in the way described by grammarians, linguists andpsycholinguitics when accepting and interpreting sentences withresumptive pronouns.

The third issue that we would like to discuss is to what extent thecharacteristics of heritage Spanish can be attributed to attrition or toincomplete acquisition. In fact, the LC Spanish may have also undergoneattrition since, the many years of contact with the L2, English, mayinfluence their L1. For instance, it has been suggested in the literaturethat even if there is not clear-cut evidence of transfer from the L2 intothe L1, there may be a convergence in that L2 speakers may preferconstructions that both languages share versus those that are specific toone of the two languages (Sánchez 2003). In this respect, the Spanish 4544


grammar of our LC group may be a somehow ‘reduced’ version of theSC group’s grammar. Since this type of convergence has also been foundin heritage speech (Montrul 2004), it may be the case that both our LCand our EC group show evidence of this type of attrition.

Finally, we would like to address the issue of whether it is attritionor incomplete acquisition that explains the characteristics of the heritagegrammar. While there is no doubt that the acquisition of Spanish was ledto completion in the case of all the members of our LC group, in the caseof our EC group, as is the case with most heritage speakers, contact withthe minority language may vary and be inconsistent. For instance, insome cases, one of the parents may be native speaker of the majoritylanguage. In other cases, children may attend kindergarten or even daycare in the majority language at a very early age, so that their use of theminority language is extremely reduced, to the point that lack of enoughexposure or usage amounts to incomplete acquisition. In the case of ourEC group, if early exposure to the language (i.e. before the critical periodfor native-like bilingual acquisition which can be set at the age of 4)2 leadsto complete acquisition, we do not have the actual individual history forthe ten speakers who were either born in Canada or arrived during thefirst year of age. However, there are two speakers who did not arrive inCanada until they were 4 years of age. In this case, if usage andprocessing of resumptive pronouns and relative qualify as late-learnedrules, these speakers might not have mastered these constructions, thustheir grammar would not differ from the grammar of the other tenheritage speakers.3 However, if they are late-learned constructions, thenthese two speakers could behave differently in that their Spanish couldhave undergone attrition as described above, in which case, theirgrammar would look more like the one of the LC group. While due tospace restrictions we will not be dealing with individual performance inthis paper, it is important to keep in mind the relevance of variability interms of individual input and of the nature of the linguistic structureswhen investigating not only the characteristics of monolingual speechbut specially those of biligual speech in general and heritage speech inparticular.

3. Resumptive pronouns in native andnon-native grammarsRelative clauses are instances of so-called long distance dependencies(LDD) because they contain phrases that have been ‘displaced’(Chomsky 1995) from their canonical position. In (1), the object ofnecesitamos has been displaced from the position after the verb, asrepresented by the t which appears in the corresponding gap.

(1) Esa es la profesora quei necesitamos ti4

That is the teacher thati we need tiIn (2), the resumptive pronoun (RP) version of (1), which is grammaticalin languages such as Arabic, the gap is filled with a resumptive pronoun.(2) *Esa es la profesora que la necesitamos

*That is the teacher that we need herThus, the first issue that we would like to point out is that in RP (orindirect) relatives, the gap is filled, which, depending on the analysis, mayconstitute an important issue in terms of the representational andprocessing cost of these constructions versus the conventional relatives in(1).

Natural languages exhibit at least two types of RPs, ‘true’ RPs as lain (2), which alternate with gaps (Kayne 1981; Sells 1984; McCloskey1990; Asudeh 2007), and ‘intrusive’ RPs as las in (3) which have beenanalyzed as instances of a ‘last resort’ or ‘repair strategy’ mechanismwhich is put to use when a trace is separated from its antecedent by anisland boundary (Shlonsky 1992), as in (3) versus (4). (3) No puedo encontrar las gafasi que siempre dejo donde lasi puedo ver

?I cannot find the glassesi that I always leave where I can see themi(4) *No puedo encontrar las gafasi que siempre dejo ti donde puedo ver ti

*I cannot find the glasses i that I always leave t i where I can see t iRPs such as las/them in (3) are considered obligatory, appear when thereis no movement of a wh-operator and can surface as grammatical optionsin any natural language. Suñer (1998) offers an account of this freealternation. She proposes that RPs are spell-outs of relative pronounswhich remain in situ. This in situ option is available in some languagesand registers and is assumed to be optional and costly. But given the factthat RPs are instances of no movement, and no movement is consideredby many to be a less marked option, what is the learnability assumptionthat we should make when comparing them to the gap equivalents whichare instances of the supposedly more marked movement option? Asudeh(2004, 2007) argues that a processing theory of ‘intrusive’ resumptivesaccounts for why gaps—not resumptives—are the preferred mechanismfor this type of dependencies. This implies that in processing terms, ‘true’RPs are also more costly than their gaps counterparts in spite of the factthat the canonical position of the displaced element is phoneticallyrealized. Therefore, if resumptives are optional, marked and costly, theymay be subject to attrition in the case of immigrant or heritage grammarsor qualify for fossilization both in the case of non-native (L2) grammarsand in the case of heritage grammars. 4746


If we look at evidence from first language (L1) acquisition, we see thatRPs are abundant mainly in experimental data from languages such asEnglish, French or Spanish where they are not sanctioned by the adultgrammar (Flynn & Lust 1981; de Villiers et al. 1990; Labelle 1996; Pérez-Leroux 1993, 1995). The profuse acceptance of RPs by L1 children(when compared to adults) in grammaticality judgments led McKee &McDaniel (2001) to propose that they are a result of a processing deficitdue to working memory limitations related to the recovery of featureswhich are no longer available in memory. This account goes along thelines of the ‘repair’ value of RPs but implies that it is only for children,not adults, that the gap alternative is more costly.

In the case of L2 acquisition, it is difficult to compare the availableresults to those obtained in the case of L1 acquisition because most of theliterature on RPs produced by L2 learners has only dealt with ‘true’ RPsand most studies have addressed the issue of transfer, and specifically thedegree of markedness, of the various relativized positions (Eckman 1977;Gass 1979; Liceras 1986) as per the typological hierarchy of markednessproposed by Keenan & Comrie (1977). More recent work by Eckmam(2004) investigates whether the presence of resumptive pronouns in non-native and near-native grammars constitutes a reflection of the speakers’processing needs. In terms of the role of input, Scott (2005) found thatL1 English speakers learning Irish had many problems incorporating‘true’ RPs into their grammar. The use of RPs in L2 seems to be ratherpervasive when it comes to transferability from the L1, and RPs do notseem to be easily incorporated into the L2 grammar when the L1 doesnot allow them. The fact that only ‘repair’ RPs are possible in bothEnglish and Spanish, and that standard Spanish optionally allows ‘true’RPs under special circumstances, may make these constructionscandidates for convergence, in other words, the degree of acceptance ofthese constructions by our LC and EC groups may differ from that ofthe SC group.

4. Restrictive relative clauses withresumptive pronouns in native Spanish:syntactic and processing accounts

Even though the previous discussion has already pointed to the factthat when dealing with relative clauses syntactic and processing issuesare very much interconnected, there are aspects of relativization thathave been approached from the representational point of view andaspects that have been clearly left to the processing field. We will firstdiscuss the competence accounts provided by linguistic theory which, inprinciple, are to be taken as proposals on how relative clauses in generaland RP relative clauses in particular are represented in the mind of the

native speaker. Besides addressing the issue of the acceptance of ‘repair’RP relatives as the one in (3) in many languages, including Spanish, wewill pay attention to RP relatives such as (5) which are ungrammatical inSpanish, versus RP relatives with indefinite antecedents such as (6) whichare considered grammatical in Spanish. (5) *Es la mujer que nunca la hemos visto llorar

It is the woman that never her we have seen cry(6) Es una mujer que nunca la hemos visto llorar

It is a woman that never her we have seen cryIn the second part of this section we will discuss the preferences shownby native and non-native speakers of Spanish when processingambiguous relative clauses such as (7) .(7) Es el hijo de la periodista que vive en Mallorca

It is the son of the journalist who lives in MallorcaThe ambiguity is based on the fact that either the first DP (el hijo), thehead, or the second DP (la periodista), the complement of the DP el hijode la periodista can be the antecedent of the relativized subject in theembedded clause. Thus, either el hijo or la periodista can live in Mallorca.

What we want to investigate is whether native, immigrant andheritage Spanish speakers have similar intuitions about ‘repair’ RPrelatives and RP relatives with indefinite antecedents and resort tosimilar strategies for the resolution of this type of ambiguity.

4.1. The syntax of restrictive relatives As we saw in the previous section, relative clauses are instances of longdistance dependencies because the relativizad element is ‘displaced’ fromits canonical position, so that, as shown in (8), the direct object positionis empty.

(8) Esa es una candidatai [CP a quieni deberíamos entrevistar____] That is a candidate whom we should interview

Sentence (8) is an instance of a direct object gap relative which, followingSuñer’s (1998) analysis, occurs when the complementizer is[+pronominal], as shown in (9).

(9) [CP una [NP candidatai [CP a quieni [C+pronominal] [IP deberíamos entrevistar a quieni]]]]

A [+pronominal] complementizer can also attract a null relative pronounto the CP position, as in (10).

(10) Esa es una candidata que deberíamos entrevistarIn this case, the relative pronoun moves to the CP position but does notundergo lexicalization, as shown in (11).

4948


(11) [DP [NP una candidatai [CP relproi [C que+pronominal] [IF deberíamos entrevistar relproi]]]]

Finally, when the complementizer is [-pronominal], Spanish directobject relatives are realized as RP relatives, as in (12).(12) Es una candidata que la deberíamos entrevistarThe difference between (10) and (12) is the presence of the resumptivepronoun la in the latter. The representation of (12) is depicted in (13),where the relative pronoun, which does not display a phonetic matrix,appears in situ, coindexed with the null operator in the CP position.(13) [DP [NP una candidatai [CP Opi [C que-pronominal] [IP lai debemos

entrevistar a quieni]]]]

According to Trujillo (1990) and also Suñer (1998) and Brucart (1999),when the antecedent is [-definite], the resumptive pronoun appears morefrequently and is more accepted than in cases when it is [+definite], as in(14), where the definite nature of the resumptive makes its presenceredundant.(14) *Es la candidata que la deberíamos entrevistar

It is the candidate that her we should interviewBrucart (1999) states that the distance between the antecedent and thecoindexed trace may favor the presence of the clitic because the cliticcontributes the gender and number features which allow for theestablishment of the relatioship between the two positions, as in (15).(15) Es la mujeri [CP[ESPECOpi][C quei][SFno sé cuando fue la última vez

que lai vimos a quieni llorar]]]

the woman that I not know when was the last timethat her we saw cryAccording to Brucart (1999), a sentence such as (16) is used morefrequently and it is more aceptable than a sentence such as (17).(16) El atracadori, a quieni algunos testigos aseguran haberloi visto [ti] por la zona anteriormente, entró en el banco a cara descubierta

The robber, whom some witnesses attest to having seen himaround the area before,

entered the bank without covering his face(17) ?? El atracadori, a quieni loi vieron [ti ] por la zona anteriormente,

entró en el banco a cara descubiertaThe robber, whom they saw him around the area before, entered the

bank without covering his faceIn (17) the relative pronoun and the trace are closer, in fact they are inthe same sentence, which makes the sentence less frequent and hardlyacceptable.

Thus, in terms of the representation of RP relatives in the grammarof native Spanish speakers, traditional grammar (Esbozo, RAE 1973)restricts them to ‘vulgar’ or ‘colloquial’ speech even though it admits thatthey are also santioned by literary usage, while linguists such as Brucart(1999) and Suñer (1998), among others, argue that RP relatives areoptionally used by native speakers when the antecedent is [-definite] orit is distant from the relativized position. The distance factor is alsorelated to the universal tendency to use resumptive pronouns, in otherwords, to the so-called ‘repair’ RP relative strategy that we described insection 3. Therefore, it is a fact that while the Spanish grammar mayreject resumptive pronouns, the parser may make use of them.

4.2. Ambiguous restrictive relatives: the role of the resumptivepronoun in forcing high and low attachment

Besides taking into consideration the two factors (definitness anddistance) which make RP relatives acceptable in native Spanish, we aregoing to address a specific issue that has been discussed in the processingliterature in order to determine whether immigrant and heritage Spanishspeakers evidence similar or different sensibility from native speakers tothe gender and number features of pronouns when processingambiguous relative clauses. As we have mentioned in the previoussection, languages display different choices when choosing an antecedentin the case of ambiguous sentences such as (7) above, and (18).

(18) [Juan me presentó a [NP [DP1 la hija j] de [DP2 la modista k]] [que[e]j/k vive en París]]]

Juan me introduced the daughter of the dressmaker thatlives in Paris

As the indexes show, the subject of the relative clause which is depictedas an empty category, can be attached (or coindexed) to DP1, la hija (theso-called high attachment strategy) or to DP2, la modista (the so-calledlow attachment strategy). In other words, there are two posible answersto the question Who lives in Paris?, one is la hija and the other one is lamodista.

It has been argued that in languages such as Dutch, French, German,Russian or Spanish there is a clear preference for high attachment whilein languages such as Arabic, Brazilian Portuguese, English, Norwegian,Romanian or Swedish, low attachment is preferred.5 High attachment isrelated to the Predicate Proximity Principle (Gibson, Pearlmutter et al.1996), according to which the relative is attached to the structure that isclosest to the predicate, in other words, to the head DP, la hija, of thecompound NP (la hija de la modista). This principle has been said to bethe one favored by languages with a relativly free word order, such as 5150


Spanish. Languages such as English are said to follow the RecencyPrinciple (Gibson and Pearlmutter 1998), which leads the parser tochoose the constituent which is closest to the relative, thus, theselanguages follow the low attachment strategy which favors the choice ofDP2, la modista.

We have shown (Senn and Liceras 2007; Senn 2008) that in directobject relatives such as (19) which displays the type of ambiguity shownin (18), when the gender and/or number feature of the two DPs isdifferent, the resumtive pronoun may resolve the ambiguity forcing lowor high attachment, as shown in (20) and (21).(19) Pintaron los techos de las casitas [que [cuando nos fuimos a viviral campo] estaban remodelando].(20) Pintaron los techos de las casitas [que [cuando nos fuimos avivir al campo] las estaban remodelando].(21) Pintaron los techos de las casitas [que [cuando nos fuimos a viviral campo] los estaban remodelando].In (20) the presence of the feminine gender feature of the resumptiveforces low attachment while in (21) the masculine gender feature forceshigh attachment.

Thus, in spite of the fact that Spanish seems to favor the highattachment strategy, so that the preferred interpretation in (19) would belos techos, that is not possible in (20), while it is the only possible choicein (21). In fact, the presence of the resumptive is welcome due to the longdistance that exists between the antecedent and the relativized position.

What we investigate here is whether possible attrition in the case ofthe immigrant group and either attrition or incomplete acquisition in thecase of the heritage group leads to insensibility to the gender and numberfeatures of the resumptive pronouns. This, together with theparticipants’ preferences with respect to the acceptance of resumptiveswith indefinite or long distance antecedents allows us to provide newdata on the gramatical competence and processing strategies of thisspecific types of bilinguals.

5. The study6

As we indicated above, and in order to contribute directly to theemerging field of heritage studies and indirectly to the fields ofbilingualism and contact linguistics, we have compared the status ofresumptive pronouns in the Spanish grammar of a group of heritagespeakers in Canada with that of a group of immigrants and a group ofnative speakers. In this section we first state the research questionsguiding our study, then provide a description of the three groups of

participants and a summary of the language background and languageassessment questionnaires that were given to all participants. We finallydescribe the two experimental tasks as well as the results.

5.1. Research questionsThe first research question that we want to answer is whether the threegroups of speakers have the same implicit knowledge of the grammar ofSpanish RP restrictive relatives. Specifically we want to find out whetherthese groups differ in their rejection of ungrammatical RP restrictiverelatives and their acceptance of RP restrictive relatives that areconsidered grammatical because they meet the indefinite or the longdistance antecedent requirement.

The second research question is whether the three groups of speakersrely on the gender and number features of the resumptive when they areasked to choose low or high attachment in ambiguous restrictive relativeclauses or whether they rely on one of the two strategies regardless of thefact that the resumptive may force the low attachment choice which,according to the processing literature, is favored by English but not bySpanish.

The specific hypotheses that we would like to test are as follows:(i) If no attrition has taken place, immigrant speakers and heritagespeakers will not differ significantly from native speakers in theirgrammatical judgments or their processing strategies.(ii) If attrition has taken place, the immigrant group and heritagegroup will differ from native speakers in their grammatical judgmentsand their processing strategies.(iii) If heritage speakers significantly differ from both the native andthe immigrant speakers, and heritage speakers also significantly differfrom native speakers, we will have to conclude that heritage speakersmay be exposed to an input that is different from the one native speakersin Spanish-speaking countries are exposed to.

5.2. Participants• Heritage or Early Contact Group This group consists of twelve English/Spanish speakers. Eight were bornin Canada or arrived in Canada before they were one year old and twoarrived in Canada at the age of 4 (the group average is shown in Table 1).Six were born to a Spanish-speaking family from Spain (the SpanishEarly Contact or EEC group) and six to a Spanish-sepaking family fromArgentina (the Argentinian Early Contact or AEC group). In some casesonly the father or the mother were immigrants from either Spain orArgentina.

5352


• Immigrant or Late Contact Group Out of the twelve participants in this group, six were born in Spain (theSpanish Late Contact or ELC group) and six in Argentina (theArgentinian Late Contact or ALC group). All of them immigrated toCanada as adults and have lived in Canada for at least ten years (theaverage time is shown in Table 1).

• Native or Spanish Control GroupThere are twenty participants in this group. Ten were born and havealways lived in Argentina (the Argentinian Spanish Control or ASCgroup) and ten were born and have always lived in Spain (the CastilianSpanish Control or ECG group).

Table 1. Summary of data from Language Background Questionnaire,SGEL and Self-evaluation Questionnaire

Spanish Argentinian Spanish Argentinian Spanish Argentinian

Control Control LC LC EC EC

# of subjects 10 10 6 6 6 6

Female/Male 10/0 8/2 5/1 1/5 4/2 4/2

Age 21.2 41.5 51.5 51.8 21 21.3

Age of arrival in

Canada N/A N/A 29.3 29.5 0.8 0.8

# years of contact

with English N/A N/A 21.18 22.3 20.1 20.5

Preferred language N/A N/A Spanish 2English English English

4Spanish

SGEL 95.7 94.7 84.6 91 80.3 81.6

%Usage Spanish N/A N/A 76.3 50.25 30.4 39.6

%Usage English 21.6 47.1 58.8 57.4

%Usage Other 1.6 2.6 4.0 3.0

Effort English N/A N/A 3.3 2.0 1.01 1.03

(1<9)

Effort Spanish 1.4 1.11 2.16 2.84

(1<9)

All participants completed a short biodata and language backgroundquestionnaire (see Appendix 1). Table 1 contains the average age for eachof the groups and subgroups, as well as general information on theaverages obtained by the different groups in the objective language

assessment questionnaire (SGEL) and the self-evaluation questionnairethat we describe in the next section.

5.3. Grammatical competence: objective and subjective assessmentAll participants completed two tests intended to provide us with anobjective measure of the participants overall competence in Spanish aswell as their subjective self-assessment of the effort they had to put intoperforming oral and written comprenshion and production activities inSpanish.

• Proficiency testA slightly modified version of the Examen clasificatorio: español paraextranjeros published by SGEL was completed by all participants. Thisis a widely used proficiency test which consists of 100 multiple choiceitems. Eighty five items out of the one hundred are administered in awritten form and fifteen in an oral form and are intended to evaluatesome of the most salient lexical, morphosintactic and semantic aspects ofstandard Spanish. We modified the fifteeen oral items to make sure thatthe test could be completed in a written form.7 Details on the results ofthis proficiency test are depicted in Table 2.

Table 2. Global proficiency (results of SGEL proficiency test)

Native Control Inmigrant Late Heritage Early ContactGroup (SC) Contact with English with English Group

Group (LC) (EC)Mean (%) 95.25 87.83 81.83Standard Deviation 2.468 8.912 9.203

An independent T-test groups for means which compares the meansof the three groups shows that the results obtained by the SC group aresignificantly different from the results obtained by the LC group (p =0.001) and the EC group (p = 0.0001). No significant differences werefound between the two experimental groups. In other words, both theimmigrant group and the heritage group significantly differ from thenative group but do not differ between themselves. This is interestingbecause it suggests that, as groups, and in terms of global competence,whether it is attrition or incomplete acquisition or, in the case of theheritage group, a low literacy level in Spanish (this test has beenspecifically designed for the actual L2 classroom) these Spanish speakersare closer to L2 intermediate and advanced learners of Spanish than to 5554


our group of native speakers. This implies that, even though there areimportant individual differences, the experimental groups could beconsidered ‘near-native’ speakers of Spanish in relation to this test (Senn2008).

• Bilingual and multilingual self-assessment questionnaireThe two experimental groups completed a self-evaluation assessmentquestionnaire intended to measure the degree of effort they felt they hadto make when producing and understanding English or Spanish ineveryday life situations. It also measured their daily contact with each ofthe two languages. The degree of effort is measured according to a 1<9scale, 1 being minimum effort and 9 maximum. A sample of thequestionnaire is provided in Appendix 28.

A summary of the self-evaluation results for Spanish obtained fromthe experimental groups is provided in Figures 1 and 2. Figure 1 showsthat the Late Contact group is very homogenous in assessing the effortinvolved in the four language skills (oral comprehension, oralproduction, written comprehension and written production) and it alsoshows that they consider using Spanish to be rather effortless.

Figure 1. Self-evaluation questionnaire. LC (Immigrant) group results.

Figure 2 shows that the Early Contact group, the group of heritagespeakers, is much less homogenous than the immigrant group and seems

to feel that a substantial amount of effort is involved mainly in relationto the written skills.

Figure 2. Self-evaluation questionnaire. EC (Heritage) group results.

Thus, when comparing the three groups global competence as measuredby the SGEL test, we see that the immigrant and the heritage group aresignificantly different from the native group and not from each other.However, in the subjective task, the heritage group is rather differentfrom the immigrant group both in terms of confidence and in terms ofhomogeneity.

5.4. Experimental tasksTwo experimental tasks were administered to all participants. Beforecarrying out the tasks, the participants read the instructions andcompleted the training exercises. The sentences were displayed in acomputer screen using the Microsof Office PowerPoint program. Eachsentence was shown for 15 seconds in the case of the Spanish Controlgroup and for 20 seconds in the case of the two experimental groups. Wefelt that given the fact that the experimental groups had less contact withthis language, they would be less used to reading in Spanish and wouldtherefore need more time to read the sentences. Time was constrained inorder to ensure that the participants did not have time to carry out ametalingistic analysis of the sentences but rather processed them on line.

5756


• Experimental Task #1: Acceptability of RP relatives This task was intended to evaluate the degree of acceptability of the RPrelatives in relation to two specific factors: (i) distance between theantecedent and the resumptive pronoun; (ii) definiteness of theantecedent. This task contained a total of 24 experimental sentences and12 filler sentences9. The four conditions tested were:(i) Four grammatical RP relative clauses: Distance. (22) Quiso que le arreglara a Ana el vestido [que [no meacuerdo [dónde lo compré]]]

He/she wanted that her I would fix to Ana the dress [that [I do notremember [where it I bought]]](ii) Four grammatical RP relative clauses: SN [- definido].(23) Marcelo tenía algunos problemas [que los había calladopor mucho tiempo]]]

Marcelo had some problems [that them he had kept silentabout for a long time]]]](iii) Eight ungrammatical RP relative clauses: No distance.(24) *Los perros odian al gato [que lo oímos maullar todo el tiempo]

the dogs hate the cat [that him we hear meowing all the time](iv) Eight grammatical gap relative clauses (25) Mario salió con una compañera del colegio que ganó una becapara estudiar en Checoslovaquia

Mario went out with a school mate who got ascholarship to study in CzechoslovakiaOnce the participants read each sentence on the screen, they had to markon a paper sheet the degree of aceptability according to a 1 to 5 scale10.

Table 3 shows the rate of acceptability of the four differentconditions. The graphic depicts the means obtained by the threedifferent groups for each condition. The 1 to 5 scale which appears onthe ordinate axis refers to the acceptability value that participants couldassign to each sentence.

An ANOVA performed on the data does not yield significant resultsamong the groups with respect to the various RP conditions. However,it indicates that there are significant differences among the groups withrespect to the the gap ralatives (p = 0.026). In fact, the EC group issignificantly different from the SC group ((p = 0.023), but the LC groupis neither significantly different from the SC group (p = 0.407) nor fromthe EC group (p = 0.394).

Therefore, what these results show is that the LC (immigrant) groupand the EC (heritage) group have the same intuitions about the RPrelatives as the SC (native) group. However, it is not the high acceptanceof RP relative clauses that unifies them but its rejection. In other words,it seems to be the case that the three groups of Spanish speakers reject

both the ungrammatical RP relatives and the RP relatives with indefiniteor long-distance antecedents that, according to grammarian andsyntacticians (Brucart 1999; Suñer 1998, among others) are bothproduced and accepted in modern standard Spanish. We can thenconclude that it is not obvious that the presence of these two factorsmake RP relatives more acceptable to native, immigrant or heritagespeakers.

Figure 3. RP relative clauses versus gap relative clauses

The heritage group seems to have problems with gap relatives,the ones that have been analyzed as having a [+pronominal] feature in theComplementizer—as we saw in examples (9) and (11) that we repeat herefor convenience.

(9) [DP una [NP candidatai [CP a quieni [C+pronominal] [IP deberíamos a quieni]]]]

(11)[DP [NP una candidatai [CP relproi [C que+pronominal] [IF deberíamos entrevistar relproi]]]]

We would like to suggest that the problems encountered by the ECgroup may be due to the optional availability of these two types of[+pronominal] relatives in Spanish. It has been shown that English L1learners of Spanish have problems abandoning their pronominal option

5958


when using or providing grammaticality judgements of Spanish subjectrelative clauses (Liceras 1986), which is due to transfer from Englishsince English has the who/that alternation11. In the case of direct objectrelatives, it may be difficult for heritage speakers to incorporate the twooptions—overt and covert relative pronoun as in (9) and (11)respectively—in their Spanish grammar because none of the fourpossible options in the English equivalent direct object relative(whom/who/that/Ø) is marked with a preposition12.

Experimental Task #2: Clitic-triggered attachmentThis task was designed to determine whether our participants would relymore on the gender and number features of the resumptive pronoun thanon the attachment strategy favored by Spanish. Thus, if we assumme thatnative speakers favor the high attachment strategy while immigrant andheritage speakers may be influenced by English and consequently haveless clear-cut intuitions with respect to the Spanish preferred attachmentstrategy, we may find that native speakers always rely on the cliticpronoun to choose one of the two potencial antecedents in an otherwiseambiguous relative clause. However, if sensitivity to the features of theresumptive pronoun is weaker than the choice of one of the twoattachment strategies, our participants will incorrectly obviate the needto choose the antecedent which is coindexed (has the same gender andnumber) with the resumptive pronoun.

Participants were shown 64 experimental RP restrictive relatives andwere asked to read the sentences and choose one of the two NounPhrases in what would be an ambiguous compound antecedent had it notbeen because the resumptive pronoun forces the choice of one of thetwo. There responses on a paper sheet. As in the case of the previousexperiment, the sentences were displayed on a computer screen using theMicrosoft Office PowerPoint program. Each sentence was shown for 15seconds in the case of the Spanish Control group and for 20 seconds inthe case of the two experimental groups. There were 32 filler sentencesthat were made up of subordinate clauses, sentences with dative andaccusative clitic-doubling, noun-noun compounds and deverbalcompounds. Participants were asked to take a break when they hadcompleted item #50.

The experimental sentences were designed on the basis of fourdifferent conditions related to both the two attachment strategies (highand low attachment) and the two factors which favor the acceptance ofRP relatives in Spanish (definiteness and distance), as follows:

• Clitic-triggered high attachment—Eight RP restrictive relatives with [+ definite] [+ distance] antecedent(26) Juan le pidió un autógrafo al domador de los leones que [ni bien llegamos al circo [lo

fotografiamos]] Juan asked for an autogrpah to the trainer of the lions that [as soon as we arrived to the circus

[him we took a picture of]]

¿A quién o a qué fotografiamos? Al domador / a los leones Whom or what did we take a picture of? The trainer / the lions

—Eight RP restrictive relatives with [- definite] [+ distance] antecedent

(27) Gloria quiere comprar una marca especial de vinos que [desde el año 2000 [la conocen en todo elmundo]]

Gloria wants to buy a special brand of wines that [since 2000 [it-fem. they know all over the world¿Qué conocen en todo el mundo? La marca / los vinosWhat do they know all over the world? The brand / the wines

—Eight RP restrictive relatives with [+ definite] [- distance] antecedent

(28) La vaca loca mordió a la hija del tambero que [la llevaron a ver al médico del pueblo]The mad cow bit the daughter of the dairy-farmer that [her they took to see the village’s doctor]

¿A quién llevaron al médico? A la hija / al tamberoWhom did they take to see the doctor? The daughter / the dairy-farmer

—Eight RP restrictive relatives with [- definite] [- distance] antecent

(29) Les presentamos a las chicas a unos sobrinos de Enrique que [los llamaron para intervenir en uncampeonato de fútbol]

We introduced to the girls some nephews of Enrique that [them they call to participate in afootball championship]¿A quién llamaron para intervenir en el campeonato? A los sobrinos / a Enrique

Whom did they call to participate in the championship The nephews / Enrique

• Clitic-triggered low attachment —Eight RP restrictive relatives with [+ definite] [+ distance] antecedent(30)Tomamos un café con la madre de las niñas que [como competían en natación [las condecoraron

varias veces]]We had coffe with the mother of the girls that [as they were competing in swimming [them they

condecorated several times]] ¿A quién o quiénes condecoraron varias veces? A la madre / a las niñasWhom they condecorated several times? The mother / the girls

—Eight RP restrictive relatives with [- definite] [+ distance] antecedent

(31)María conoce a la cómplice de un ladrón que [algunos vecinos de mucha confianza [lo han vistopor la zona anteriormente]]

María knows the accomplice[femenine] of a thief that [some neighbours well known [him they have

seen around the area before]]¿A quién han visto por la zona? A la cómplice / al ladrón Whom have they seen around the area? The accomplice / the thief

—Eight RP restrictive relatives with [+ definite] [- distance] antecedent

(32) El granizo nos destrozó las ventanas de la habitación que [la habíamos dejado impecable antes desalir de viaje]

The hail us destroyed the windows of the room that [it we had left impeccable before going on atrip¿Qué habíamos dejado impeccable? Las ventanas / la habitaciónWhat have we left impeccable? The windows / the room

—Eight RP restrictive relatives with [- definite] [- distance] antecent 6160


(33) Carmen trabaja con el cuñado de una vecina que [la encontramos a veces en la peluquería de la esquina]Carmen works with the brother in law of a neighbour that [her we find sometimes at thehairdresser of the corner]

¿A quién encontramos en la peluquería? Al cuñado / a la vecinaWhom did we find at the hairdresser? The brother in law / the neighbour

5.5. ResultsThe results for the High Attachment condition are summarized inGraphic 4. Even though it can be observed that the overall pattern is thesame for all three groups in that all participants had the most problemswith the [+definite] [+distance] condition, the statistical analysisindicates that the EC group differs significantly from the SC group (p =< 0.0001) and also from the LC group ((p = 0.005). There are nostatistically significant differences between the SC and the LC group (p= 0.360), which means that the immigrant group does not differ from thenative group in terms of their sensitivity to the triggering effect of theresumptive pronouns.

Figure 4. Clitic-triggered Choice of High Attachment

The results for the Low Attachment condition are shown in Graphic 5.While the pattern is less homogenous than in the case of the HighAttachment condition, the statistical analysis indicates that there are no

significant differences between the SC and the LC group (p = 0.109), norbetween the SC and the EC group (p = 0.430).

Even though there were no statistically significant differences amongthe four factors, in the case of the SC group, the factor that got the largestnumber of errors was the [+definite] [-distance], which is expectedbecause none of the two factors which may contribute to theacceptability of resumptive pronouns is present. This was also the casefor the LC and the EC group but to a lower degree.

An ANOVA performed to compare the SC group and the twoexperimental groups in terms of the number of ‘correct’ (choice ofantecedent that agrees with the resumptive pronoun) responses for thefour factors shows that there is a significant difference between them (p= 0.004). What is important is that [+definite] + [+distance] antecedentsare less problematic in terms of producing errors than [+definite] + [-distance] antecedentes. In fact, a two sample t-test shows that these twofactors are significantly different in that less errors were made in the caseof the former (p = 0.049).

Figure 5. Clitic-triggered Choice of Low Attachment

It is also important to notice that the number of errors between thesentences where the clitic triggered high attachment and the sentences

6362

Linguistic theory and the analysis of minority languages

where it trigerred low attachment is not significant, which indicates thatthe resumptive can neutralize the possible preference for high or lowattachment in the case of the three groups of subjects.

6. Discussion and conclusionsBased on the results obtained from the proficiency test we can infer thatthere is a difference between the native group on the one hand and theimmigrant and the heritage group on the other. It is not clear whetherthis is so because the experimental groups contact with Spanish isreduced or because they have had less access to literary tests than nativespeakers who live in the Spanish-speaking countries (the test containsnot only items intended to measure morphosyntactic and semanticknowledge but also lexical knowledge) or because of both. Therefore, wecan conclude that the input the heritage group is exposed to is somehoweroded. The self-assesment questionnaire provides a different picture:the immigrant group is more homogenous and more confident than theheritage group when assessing its linguistic capability in Spanish. Thiscould lead us to conclude that both the immigrant and the heritagespeakers, but mainly the heritage speakers could be considered ‘near-native speakers’ (Senn 2008) even though they would belong to adifferent group than the one for which the term has been coined: L2speakers whose first contact with the second language begun afteradolescence. However, is it also the case that the immigrant group andthe heritage group are significantly different from the native group withrespect to their intuitions about the grammaticality and thedisambiguating effects of resumptive pronouns in Spanish?

This leads us to address the two research questions and the threehypotheses that we set up to answer and test repectively.

With respect to the question of whether the three groups of speakershave the same implicit knowledge of the grammar of Spanish RPrestrictive relatives, we have to answer that they do. However, based onthe results of experiment #1, their knowledge differs from the one thatgrammarians and syntactitians predicate of Spanish in that ourparticipants’ rejection of RP relatives does not differentiate in asignificant way between RP relatives with indefinite and long distanceantecedents from those with definite and short distance ones. Thus, wehave to conclude that the fact that resumptive pronouns are costly,marked and optional leads to their rejection in the type of experimentaltask that we carried out. These results confirm that the RP relativizationstrategy is rather marginal in the grammar of Modern Spanish and

Juana M. Liceras y Cristina Senn

confirm the costly, marked and optional nature of resumptive pronounsand the fact that their main role is that of ‘repair’ or ‘last resort’ strategy.

As for our second research question of whether the gender andnumber features of the resumptive determine the choice of antecedentregardless of whether they force low or high attachment in ambiguousrestrictive relative clauses, the answer is that all three groups are sensitiveto the gender and number features of the resumptive, though the heritagegroup produces significantly more errors of agreement than theimmigrant and the native group both in the High and the Lowattachment condition and significantly more in the case of the former.This may be due to influence from English, their dominant language,since having to choose high attachment is at odds with the processingstrategy favored by English (the low attachment strategy). Theprocessing role of resumptive pronouns seems to show here more thanin the representation task because the native group producessignificantly less errors with the factor combination [+definite] /[+distance] than with the factor combination [+definite] / [-distance]. Inother words, the native group has strong intuitions about theungrammaticality of ‘true’ resumptives in Spanish.

With respect to our first hypothesis, which stated that if no attritionhad taken place, immigrant speakers and heritage speakers would notdiffer significantly from native speakers in their grammatical judgmentsor their processing strategies, we have to say that it is confirmed in thatresumptive pronouns are equally rejected by all groups and there is apreference for gap relatives. However, in terms of the processingstrategies, the heritage group –be it due to attrition, incompleteacquisition, lack of literacy skills or reduced input (the immigrant groupis significantly different from the native group when it comes to overallcompetence)—is significantly different from the inmigrant and thenative group.

The second hypothesis stated that if attrition had taken place, theimmigrant group and the heritage group would differ from the nativegroup in their grammatical judgments and their processing strategies. Wehave already stated that only the heritage group differs from the othertwo with respect to the processing strategies since their sensitivity to thetriggering force of the gender and number features is not up to nativestandards. Since they do not differ from the native and immigrant groupwith respect to their rejection of resumptive pronouns, it looks as if wehave isolated an area of competence which is vulnerable, mainly whenlinked to processing strategies. In other words, while these learners maynot have problems with gender and number agreement in lesschallenging contexts (i.e.: adjective/noun agreement), they seem to lack 6564


the sensibility displayed by both the native and the immigrant speakerswhen confronted to the desambiguating role of resumptives.

Our third hypotheses stated that if heritage speakers significantlydiffered from both the native and the immigrant speakers, and heritagespeakers also significantly differed from native speakers, we would haveto conclude that heritage speakers may be exposed to an input that isdifferent from the one native speakers in Spanish-speaking countries areexposed to. This hypothesis is partially confirmed in that the immigrantgroup significantly differs from the native group in terms of the overallcompetence of the participants as measured by the SGEL test but it is notconfirmed in terms of their local competence, since they do not differfrom the native group in their preference for gap relatives and theirsensitivity to the gender and number features which lead to choose thecoindexed DP in the compound antecedents. However, they make moreerrors than the native group and they are less sensitive to the [+definite]/ [-distance] combination, which seems to suggest that there is a subtledistance from the native speakers which in turn may translate inproviding the heritage group with a reduced input.

While there are many questions that we have not answered, be itbecause of limitations to the number of issues that we can deal with hereor because our data does not allow us to answer them, we hope to havecontributed to the field of heritage studies by discussing theoretical andmethodological issues that have not been addressed before or have notbeen addressed in this specific manner. Specifically, we have shown: (i)that we need to compare heritage speakers to immigrant speakers inorder to determine what are the characteristics of heritage speech andwhether heritage speakers are exposed to reduced input; (ii) that we needto differentiate overall competence as measured by proficiency testsfrom local competence as measured by intuitions related to ‘fuzzy’ areasof the grammar such as the usage and acceptability of resumptivepronouns; and (iii) that we also need to differentiate overall competenceand grammatical intuitions from sensitivity to language specific featuresand processing strategies such as the gender and number featuresresponsible for clitic-triggered low and high attachment strategies.

To conclude we would like to stress the need to investigate the statusof minority languages in immigrant and heritage populations not onlybecause, as basic research, this type of study enhances the fields ofbilingualism and contact linguistics but also because it may provideinformation that can be applied to educational, sociological and evenpolitical endevours.

Juana M. Liceras Cristina SennDept. of Modern Languages Faculty of Creative and Critical Studies & Dept. of Linguistics Spanish ProgramUniversity of Ottawa University of British Columbia – Okanagan70 Laurier East 3333 University Way Ottawa, Ontario K1N 6N5 (Canada) Kelowna, British Columbia V1V 1V7 (Canada)Telephone: 613-562-5800, ext. 3742 Telephone: 250-807-8446Fax: (613) 562-5138 Fax: (250) 807-9900 E-mail: [email protected] E-mail: [email protected]

Notes1 These data was collected to investigate not only the status of RP relatives but also the

status of optional and obligatory clitic-doubling in native, immigrant and heritagegrammars. The second syntactic property is the one that has a different status in the twovarieties of Spanish, mainly because it is in the Rioplatense variety where direct objectclitic doubling (lo vi a Pedro) is systematically used and accepted.

2 According to the position taken, for instance, by Meisel (2008), who distinguisheslanguage acquisition from birth from any form of acquisition that occurs after the age of4, child L2A is different from bilingual first language acquisition and closer to adultsecond language acquisition. We take this as evidence that native acquisition is completedby age 4.

3 Certain control phenomena have been described as late-learned rules, that is, grammaticalphenomena for which adult native speaker ability is not attained by children until theyare six or older (Goodluck & Birch, 1987). Cromer (1987) shows that constructions suchas (This paperi is tough [PROi to finish ti]) develop late in children. Goodluck and Behne(1992) show that young children do not understand constructions such as (Daisyi choosesPlutoj [PROi to read to tj]), giving fairly random responses until a fairly late age. Tavakolian(1981) and Goodluck (1987), among others, show that young children interpret theinfinitival clause as having a sentence-external referent in constructions such as ([PROi tokiss the duck] would make the lioni very happy). Also, the development of the ability toproduce and understand complex sentences by child learners has been considered to beone of the most interesting and important aspects of language acquisition, as a distinctivetrait of human languages (Limber, 1973).

4 These [+human] direct object restrictive relatives where the complementizer position isoccupied by the complementizer que rather than by a relative pronoun (see analysis insection 4.1) are different from relatives which display a que without preposition, namelythose relatives where the verbal preposition has been omitted as in Ese es el profesor quete hablé ayer. This type of omission, as well as omission of the preposition with adverbialrelatives as in Era un río que se iba a buscar agua para tomar o Hubo un momento queellos se fueron con el papá, seems to be rather frequent in colloquial speech (Navarro2006). According to this author, the rate of this type of omission is not affected by socialfactors such as age or gender, although there seems to be a higher frequency of omissionamong younger speakers, in general, and younger speakers from less educated families,in particular.

5 See Cuetos and Mitchell (1988), Guillelmon and Grosjean (2001), Papadopoulou andClahsen, (2003), Fernández (2003), Dussias (2004), Gibson, Pearlmutter and Torrens(1999), Hemforth, Konieczny and Scheepers (2000), Brysbaert and Mitchell (1996),Zagar, Pynte and Rativeau (1997), Ehrlich, Fernandez, Fodor, Stenshoel and Vinereanu(1999), Miyamoto (1998) or Abdelghany and Fodor (1999), among others).

6 This study was conducted with the approval of the Research Ethics Board of theUniversity of Ottawa (Ethics file # 05-06-11; approved June 12, 2006).

7 Sánchez, Aquilino and Terencio Simón. 1980. Examen clasificatorio: español para 6766


extranjeros. Madrid: Sociedad General Española de Librería (SGEL). 8 J. M. Liceras, A. Tremblay and R. Pérez-Tattam (2001), Language Acquisition

Laboratory [http://www.modernlanguages.uottawa.ca/lab.html], University of Ottawa.9 The twelve filler sentences were included to ensure that having a fixed pattern of

sentences would not influence the participants’s responses. The fillers consisted ofsubordinate clauses and clauses with clitic-doubling of direct and indirect objects.

10The scale was presented as follows:

1 2 3 4 5Suena muy mal suena mal suena raro no suena mal suena bienSounds very bad sounds bad sounds strange doesn’t sound bad sounds o.k.

11Advanced English speaking learners of Spanish continue to produce and acceptungrammatical restrictive relatives such as *una candidata quien vive en París instead ofuna candidada que vive en París, which seems to indicate that the impossibility tophonetically realize the relative pronoun in these constructions is not capture by thesespeakers.

12In fact, Montrul (2004, 2008) has found that heritage speakers have problems with the ‘a’marking of the Spanish [+animate, +specific] direct objects.

ReferencesAbdelghany, H., and J. D. Fodor. 1999. “Low attachment of relative clauses in Arabic”.

Poster presented at AMLaP (Architectures and Mechanisms of LanguageProcessing) ’99, Edinburgh, UK.

Asudeh, Ash. 2004. Resumption as Resource Management. Unpublished Ph.D. dissertation,Stanford University.

Asudeh, Ash. 2007. “Three kinds of resumption”. Paper presented at the InternationalWorkshop on Resumptive Pronouns at the Interfaces, University of Paris 7,France, June 21-22.

Birdsong, David. 1999. Second Language Acquisition and the Critical Period Hypothesis.Mahwah, NJ: Lawrence Erlbaum Associates.

Brysbaert, M. and D. C. Mitchell. 1996. “Modifier attachment in sentence parsing:Evidence from Dutch”. The Quarterly Journal of Experimental Psychology49A (3): 664-695.

Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: The MIT Press.Cromer, Richard. F. 1987. “Language growth with experience without feedback”. Journal

of Psycholinguistic Research, 16, 223–231.Cuetos, F., and D. C. Mitchell. 1988. “Cross-linguistic differences in parsing: restrictions

on the use of the Late Closure strategy in Spanish”. Cognition 30: 73-103.De Villiers, Jean, Thomas Roeper and Anne Vainikka. 1990. “The acquisition of long-dis-

tance rules”. In Language Processing and Language Acquisition, L. Frazier and J.De Villiers (Eds.), 257-297. Kluwer Academic Publishers.

Dussias, P. E. 2004. “Parsing a first language like a second: The erosion of L1 parsing stra-tegies in Spanish-English bilinguals”. International Journal of Bilingualism 3:355-371.

Eckman, F. 1997. “Markedness and the contrastive analysis hypothesis”. LanguageLearning 27: 315-330.

Eckman, F. 2004. “Optimality Theory, markedness, and second language syntax: The caseof resumptive pronouns in relative clauses”. Studies in Phonetics, Phonology,

Morphology 10: 89-110.Ehrlich, K., E. Fernández, J. D. Fodor, E, Stenshoel and M. Vinereanu. 1999. “Low

attachment of relative clauses: New data from Swedish, Norwegian andRomanian”. Paper presented at the 12th Annual CUNY conference on HumanSentence Processing, New York.

Fernández, E. M. 2003. Bilingual sentence processing: relative clause attachment in Englishand Spanish. Amsterdam/Philadelphia: J. Benjamins.

Flynn, Suzanne and Barbara Lust. 1981. “Acquisition of relative clauses: developmentalchanges in their heads”. In W. Harbert and J. Herchensohn (Eds.). CornellWorking Papers 1: 33-45. Ithaca, NY: Cornell University.

Gass, S. 1979. “Language transfer and universal grammatical relations”. LanguageLearning 29: 237-344.

Gibson, E. and N. Pearlmutter, N. 1998. “Constraints on sentence comprehension”.Trends in Cognitive Science 2: 262-268.

Gibson, E., N. Pearlmutter, N. Canseco-Gonzalez and G. Hickok. 1996. “Recency pre-ference in the human sentence processing mechanism”. Cognition 59: 23-59.

Gibson, E., N. Pearlmutter, N. and V. Torrens. 1999. “Recency and lexical preferences inSpanish”. Memory and Cognition, 27, 603-611.

Goodluck, Helen and Dawn Behne. 1992. “Development in control and extraction”. InTheoretical studies in language acquisition: Papers from the Berlin workshop, J.Weissenborn, H. Goodluck and T. Roeper (eds.), 153–173. Hillsdale, NJ:Lawrence Erlbaum.

Goodluck, Helen, and Barbara Birch. 1987. “Late-learned Rules in First and SecondLanguage Acquisition”. Unpublished manuscript.

Goodluck, Helen. 1987. “Children’s interpretation of pronouns and null NPs: An alter-native view.” In Studies in the acquisition of anaphora, Vol. II. B. Lust (ed.),247–269. Dordrecht, Netherlands: Reidel.

Guillelmon, D. and F. Grosjean. 2001. « The gender marking effect in spoken word recog-nition: The case of bilinguals”. Memory and Cognition 29(3): 505-511.

Hemforth, B., L. Konieczny and C. Scheepers. 2000. “Modifier attachment: relative clau-ses and coordinations”. In German sentence processing, B. Hemforth and L.Konieczny (eds.), 159-183. Dordrecht, NL: Kluwer Academic Press.

Johnson, Jacqueline and Elisa Newport. 1991. “Critical period effects on universal prin-ciples of language. The status of subjacency in the acquisition of a second lan-guage”. Cognition 39, 215–258.

Kayne, R. 1981. “On some differences between French and English”. Linguistic Inquiry12: 349-371.

Keenan, E. and B. Comrie. 1977. “Noun phrase accessibility and universal grammar”.Linguistic Inquiry 8:63-99.

Labelle, M. 1996. “The acquisition of relative clauses: movement or no movement?”Language acquisition 5: 65-82.

Lenneberg, Eric. 1967. Biological Foundations of Language. New York NY: Wiley.Liceras, Juana. 1986. Linguistic Theory and Second Language Acquisition: the Spanish Non-

native Grammar of English Speakers. Tübingen: Gunter Narr. Limber, John. 1973. “The genesis of complex sentences”. In Cognitive development and the

acquisition of language, T. Moore (ed.), 169–186. New York: Academic Press.McCloskey, James. 1990. “Resumptive pronouns, A’-binding and levels of representation

in Irish”. In The Syntax and Semantics of Modern Celtic Languages. Syntax andSemantics 23, R. Hendrick (ed.). New York: Academic Press.

McKee, Cecile and D. McDaniel. 2001. “Resumptive pronouns in English relative clau-ses”. Language Acquisition, 9, 113-156.

Meisel, Jürgen. 2008. “Child second language acquisition or successive first languageacquisition?” In Current trends in child second language acquisition, B. Haznedar 6968


and E. Gavruseva (eds.), 55-80. Amsterdam: John Benjamins.Miyamoto, E. T. 1998. “Relative clause attachment in Brazilian Portuguese”.

Unpublished manuscript, Massachusetts Institute of Technology, Cambridge.Montrul, Silvina. 2004. “Subject and object expression in Spanish heritage speakers: A case

of morpho-syntactic convergence”. Bilingualism: Language and Cognition, 7(2),125–142.

Montrul, Silvina. 2005. “Second language acquisition and first language loss in adult earlybilinguals: Exploring some differences and similarities”. Second LanguageResearch, 21(3), 199–249.

Montrul, Silvina. 2008. Incomplete Acquisition in Bilingualism. Re-examining the Age Factor.Amsterdam: John Benjamins.

Navarro, Manuel. 2006. “La frecuencia de los relativos en el habla de Valencia(Venezuela). Boletin de Lingüística: 18(25), 66-99. Available in World WideWeb: <http://www.scielo.org.ve/scielo.php?script=sci_arttext&pid=S0798-97092006000100003&lng=es&nrm=iso>. ISSN 0798-9709.

Papadopoulou, D., and Harald Clahsen. 2003. “Parsing strategies in L1 and L2 sentenceprocessing: A study of relative clause attachment in Greek”. Studies in SecondLanguage Acquisition 25: 501-528.

Penfield, Wilder. 1953. “A consideration of the neurophysiological mechanisms of speechand some educational consequences”. Proceedings of the American Academy ofArts and Sciences. 82, 199–214.

Penfield, Wilder and Lamar Roberts. 1959. Speech and Brain Mechanisms. Princeton NJ:Princeton University Press.

Pérez-Leroux, Ana Teresa. 1993. Empty Categories and the Acquisition of Wh-movement.Unpublished Ph.D. dissertation, University of Massachusetts.

Pérez-Leroux, Ana Teresa. 1995. Resumptives in the acquisition of relative clauses.Language Acquisition 4: 105-138.

Pires, Acrisio and Jason Rothman. (2009). “Disentangling Contributing Variables toIncomplete Acquisition Competence Outcomes: What Differences AcrossBrazilian and European Portuguese Heritage Speakers Tell Us”. InternationalJournal of Bilingualism, 13.

Polinsky, Maria. 1997. “American Russian: Language loss meets language acquisition”. InFormal approaches to Slavic linguistics, W. Browne et al. (eds.), 370-407). AnnArbor: Michigan Slavic Publications.

Polinsky, Maria. 2006. “Incomplete acquisition: American Russian”. Journal of SlavicLinguistics 14, 191-262.

RAE. 1973. Esbozo de una nueva gramática de la lengua española. Madrid: Espasa-Calpe. Rothman, Jason. 2007. “Heritage speaker competence differences, language change, and

input type: Inflected infinitives in Heritage Brazilian Portuguese”. TheInternational Journal of Bilingualism 11, 359-389.

Sánchez, Aquilino and Terencio Simón. 1989. Test clasificatorio de español para extranje-ros. Madrid: Sociedad General Española de Librería (SGEL).

Sánchez, Liliana. 2003. Quechua-Spanish Bilingualism: Interference and Convergence inFunctional Categories. Amsterda/Philadelphia: John Benjamins.

Scott, Sheila. 2005. The Second Language Learning of Irish Relative Clauses: theMorphology/Syntax Interface. Unpublished Ph. D. dissertation, University ofOttawa.

Sells, Peter. 1984. Syntax and Semantics of Resumptive Pronouns. Unpublished Ph.D. disser-tation, University of Massachusetts, Amherst.

Senn, Cristina and Juana M. Liceras. 2007. “Early and late bilin gualism: resumptive pro-nouns in adult L2 Spanish”. Internacional Simposium on Bilingualism (IBS6),University of Hamburg, Hamburg, May 30 – June 2.

Senn, Cristina Rita. 2008. Reasuntivos y doblado del clítico: en torno a la caracterización deltérmino ‘casi nativo’. Unpbulished Ph. D. dissertation, University of Ottawa.

Shlonsky, Ur. 1992. “Resumptive pronouns as a last resort”. Linguistic Inquiry 25:443-468.Silva-Corvalán, Carmen. 1994. Language Contact and Change, Spanish in Los Angeles.

Oxford: Oxford University Press.Suñer, Margarita. 1998. “Resumptive restrictive relatives: a cross-linguistic perspective”.

Language 74(2): 335-363.Tavakolian, S. L. 1981. The conjoined clause analysis of relative clauses. In Language

acquisition and linguistic theory, S. L. Tavakolian (ed.), 167–187. Cambridge,MA: MIT Press.

Zagar, D., J. Pynte and S. Rativeau. 1997. “Evidence for early-closure attachment on first-pass reading times in French”. The Quarterly Journal of ExperimentalPsychology, 50A: 421-438.

7170


Appendix #1. Language Background Questionnaire

Language Background QuestionnaireUniversity of Ottawa • Department of Modern Languages and Literature

Language Acquisition Lab (LA-LAB) • Spanish Program

Course: Date: 1. Name: 2. Gender: F [ ] M [ ]3. Age: 4. Place of birth: 5. Mother tongue: 6. Mother’s dominant language:7. Father’s dominant language: 8. Language(s) spoken at home as a child:9. Language(s) you spoke during the first five years of your life: 10. Language(s) studied in (please include Spanish):

•Primary school: •Secondary (high) school: •University: •Other institutions:

11. Languages you use:•At home:•At school:•At work: •When you dream:

12. Other languages you can:•Read: •Speak: •Write:

13. What language do you feel most comfortable with at this time?14. What program are you in at the university?15. Year at the university:16. Are you presently studying Spanish at the university level? If so, please give

us university name and course number17. Why have you chosen to learn Spanish?18. Contact with Spanish outside classroom:

Present contact:•Approximate hours/week: •Context: (e.g. friends, family, clubs, etc.):Previous contact: •Have you ever visited a Spanish-speaking country? YES [ ] NO [ ]IF, YES •When? •For how long?

APPENDIX #2. FLUENCY ASSESSMENT QUESTIONNAIRE FOR BILINGUALS AND POLYGLOTS

Fluency Assessment Questionnaire for Bilinguals and Polyglots

University of Ottawa • Department of Modern Languages and Literature

Language Acquisition Lab (LA-LAB) • Spanish Program

Section 1

General information

Please indicate :

Your name ———————————————————————

Your country of origin ———————————————————————

When applicable, indicate at what age you started to :

Please indicate your language(s) of instruction through primary and secondary schoolby grades.

Section 2Use of English, Spanish and other languages

We now ask you to partition, in percentages, the time you spend using English, Spanish

or other languages in various contexts. Please make sure the sum of the percentages you

report add up to 100%. Specify the other languages you use.

When you talk with your mother

When you talk with your father

When you talk with your friends at the university

When you talk with your friends outside the university

When you talk with your colleagues at work (if applicable)

KG 1 2 3 4 5 6 7 8 9 10 11 12 13

English

Spanish

Other:Specify

In English In Spanish In another language: specify

Speak

Read

Write

7372


Section 3Effort required to carry out different activities in English or Spanish

We now ask you to evaluate, on a 9-point scale, the level of difficulty you would expe-rience if you were to carry out various activities in English or Spanish. On this scale, 1means that the use of the specified language in the specified activity would be very easyfor you. At the other end of the scale, 9 means that the use of the specified language in thespecified activity would be very difficult for you. The values 2 through 8 indicate inter-mediate levels of difficulty. In the use of this scale the sum of the two values for Englishand Spanish for an activity can vary between 3 and 27.Research on bilingualism shows that people seldom have the exact same level of fluencyin their two or more languages. Usually, one language is dominant relative to the others.Moreover, the dominant language may sometimes differ for different activities. Please usethe 9-point scale carefully so as to represent as accurately as you can the level of difficultyyou would experience if you were to carry out the following activities in English orSpanish. Please indicate your assessment in the provided spaces

..........

..........

Oral Production

9. Count rapidly from 1 to 20

10. Recite the alphabet quickly from A to Z

Activity In English In Spanish

Oral Comprehension

1. Tell the difference between a question and a request

2. Tell what the weather will be after listening to a weather forecast on the radio

Veryeasy

1 2 3 4 5 6 7 8 9 Verydifficult

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

English

Spanish

Other

..........

..........

Writing

31. Writing a postcard to a friend

32. Writing a note for someone

Reading Comprehension

23. Understanding the menu in an ordinary restaurant

24. Follow a recipe in the preparation of a dish

75

Políticas lingüísticas y deintegración en materia deemigración en Europa comoreflejo de la construcción de losestados-nación

Mercè Pujol Berché 1

Este artículo pone de relieve que los movimientos migratorios sonuna constante en la historia de la humanidad. La emigración se pre-senta como politemporal, poliespacial, polivocal y polifónica. El artí-culo se divide en cuatro partes. En la primera, se presenta un panora-ma histórico de Europa, atendiendo al concepto de ciudadano. Lasegunda parte se dedica a la emigración europea que fue a América.La tercera se dedica a la importancia de la industrialización en Euro-pa para la llegada de trabajadores procedentes de zonas rurales y depaíses limítrofes por motivos tanto demográficos, como económicos.Finalmente, se pone de manifiesto que el plurilingüismo es una carac-terística fundamental del panorama internacional y se hacen algunasreflexiones sobre cómo desarrollarse dentro del respeto a la diversi-dad lingüística.

This article explains how the migratory movements are a constant inthe humanity history. Migration appears like a poly-temporal, poly-spatial, poly-vocal and poly-phonic phenomenon. The article is divi-ded in four parts. First of all, a historical panorama of Europe appe-ars, taking care of the concept of citizen. The second part is dedica-ted to the European emigration to America. A third part is dedicatedto the importance of industrialization in Europe for the arrival ofworkers coming from countryside and bordering countries, bydemographic as well as economic reasons. Finally, it is shown that themultilingualism is a fundamental characteristic of the current inter-national panorama and some reflections are made on the develop-ment of linguistic diversity.

Palabras claves: inmigración, política lingüística, integración, Euro-pa, estado, nación

Lengua y migración 1:1 (2009), 75-120ISSN: 1889-5425. © Universidad de Alcalá

linguistic theory and the analysis of minority languages: native

Documents